def dist_exp(f_path, mutation_rate, crossover_rate, w_social):
runs = 5
data = {}
pop_size = 500
num_gens = 2000
for gc in GAMES_CODES:
print(f'\n\n Game code = {gc}')
allrows = []
for pth, best in TEST_SET_1[:2]:
row = []
print('>> Instance', instance_name(pth))
for run in range(runs):
print(f'>> Run{run+1}/{runs}')
vrp = VRP(*read_file(os.path.abspath(pth)))
game = gameClassFactory(gc)
result, routes, strategy_dist = ga_social_interaction_vrp(
vrp,
game,
pop_size,
num_gens,
mutation_rate=mutation_rate,
crossover_rate=crossover_rate,
wgt_solution=1 - w_social,
wgt_social=w_social)
row = row + strategy_dist
allrows.append(row)
with open(f'results/dists/{gc}_dist_exp.csv', 'w') as file:
writer = csv.writer(file)
writer.writerows(allrows)
def weights_exp(f_path, mutation_rate, crossover_rate, game_code):
runs = 10
data = {}
pop_size = 500
num_gens = 2000
all_data = []
header = ['solution fitness weight', 'social fitness weight'] + list(
map(lambda x: "% Gap in " + instance_name(x[0]),
TEST_SET_1)) + ['Avg % Gap']
all_data.append(header)
print(header)
for n in range(1, 12):
w_social = n * 0.05
w_solution = 1 - w_social
print(f'\n\n solution wgt = {w_solution}; social wgt = {w_social}')
row = [w_solution, w_social]
for pth, best in TEST_SET_1:
print('>> Instance', instance_name(pth))
instance_results = []
for run in range(runs):
print(f'>> Run{run+1}/{runs}')
vrp = VRP(*read_file(os.path.abspath(pth)))
game = gameClassFactory(game_code)
result, routes, strategy_dist = ga_social_interaction_vrp(
vrp,
game,
pop_size,
num_gens,
mutation_rate=mutation_rate,
crossover_rate=crossover_rate,
wgt_solution=w_solution,
wgt_social=w_social)
instance_results.append(result)
row.append(gap_percentage(mean(instance_results), best))
row.append(mean(row[2:]))
print(row)
all_data.append(row)
with open('results/weights_exp.csv', 'w') as file:
writer = csv.writer(file)
writer.writerows(all_data)
def games_exp(f_path, mutation_rate, crossover_rate):
runs = 10
data = {}
pop_size = 500
num_gens = 2000
all_data = []
header = ['game'] + list(
map(lambda x: "% Gap in " + instance_name(x[0]),
TEST_SET_1)) + ['Avg % Gap']
all_data.append(header)
print(header)
for gc in GAMES_CODES:
print(f'\n\n Game code = {gc}')
row = [gc]
for pth, best in TEST_SET_1:
print('>> Instance', instance_name(pth))
instance_results = []
for run in range(runs):
print(f'>> Run{run+1}/{runs}')
vrp = VRP(*read_file(os.path.abspath(pth)))
game = gameClassFactory(gc)
result, routes, strategy_dist = ga_social_interaction_vrp(
vrp,
game,
pop_size,
num_gens,
mutation_rate=mutation_rate,
crossover_rate=crossover_rate,
wgt_solution=0.9,
wgt_social=0.1)
instance_results.append(result)
row.append(gap_percentage(mean(instance_results), best))
row.append(mean(row[2:]))
print(row)
all_data.append(row)
with open('results/games_exp.csv', 'w') as file:
writer = csv.writer(file)
writer.writerows(all_data)
def basic_params_exp(f_path):
runs = 10
data = {}
pop_size = 500
num_gens = 2000
all_data = []
header = ['mutation rate', 'crossover rate'] + list(
map(lambda x: "% Gap in " + instance_name(x[0]),
TEST_SET_1)) + ['Avg % Gap']
all_data.append(header)
print(header)
for mr, cr in product([0.2, 0.4, 0.6, 0.8], [0.2, 0.4, 0.6, 0.8]):
print(f'\n\n mutation rate = {mr}; crossover rate = {cr}')
row = [mr, cr]
for pth, best in TEST_SET_1:
print('>> Instance', instance_name(pth))
instance_results = []
for run in range(runs):
print(f'>> Run{run+1}/{runs}')
vrp = VRP(*read_file(os.path.abspath(pth)))
game = Game()
result, routes, strategy_dist = ga_social_interaction_vrp(
vrp,
game,
pop_size,
num_gens,
mutation_rate=mr,
crossover_rate=cr,
wgt_solution=1,
wgt_social=0)
instance_results.append(result)
row.append(gap_percentage(mean(instance_results), best))
row.append(mean(row[2:]))
print(row)
all_data.append(row)
with open('results/basic_params_exp.csv', 'w') as file:
writer = csv.writer(file)
writer.writerows(all_data)
def comp_exp(f_path, mutation_rate, crossover_rate, game_code, w_social):
runs = 10
data = {}
pop_size = 500
num_gens = 2000
all_data = []
header = ['Algorithm'] + list(
map(lambda x: "% Gap in " + instance_name(x[0]),
TEST_SET_1)) + ['Avg % Gap']
all_data.append(header)
print(header)
row = ['GASI']
for pth, best in TEST_SET_1:
print('>> Instance', instance_name(pth))
instance_results = []
for run in range(runs):
print(f'>> Run{run+1}/{runs}')
vrp = VRP(*read_file(os.path.abspath(pth)))
game = gameClassFactory(game_code)
result, routes, strategy_dist = ga_social_interaction_vrp(
vrp,
game,
pop_size,
num_gens,
mutation_rate=mutation_rate,
crossover_rate=crossover_rate,
wgt_solution=1 - w_social,
wgt_social=w_social)
instance_results.append(result)
row.append(gap_percentage(mean(instance_results), best))
row.append(mean(row[2:]))
all_data.append(row)
print(row)
row = ['GA']
for pth, best in TEST_SET_1:
print('>> Instance', instance_name(pth))
instance_results = []
for run in range(runs):
print(f'>> Run{run+1}/{runs}')
vrp = VRP(*read_file(os.path.abspath(pth)))
game = gameClassFactory(game_code)
result, routes, strategy_dist = ga_social_interaction_vrp(
vrp,
Game(),
pop_size,
num_gens,
mutation_rate=mutation_rate,
crossover_rate=crossover_rate,
wgt_solution=1,
wgt_social=0)
instance_results.append(result)
row.append(gap_percentage(mean(instance_results), best))
row.append(mean(row[2:]))
all_data.append(row)
print(row)
with open('results/comp_exp.csv', 'w') as file:
writer = csv.writer(file)
writer.writerows(all_data)